Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add koolamusic/claudefiles --skill localframegit clone --depth 1 https://github.com/koolamusic/claudefilesWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/koolamusic/claudefiles/localframe)<a href="https://agentmods.dev/skills/koolamusic/claudefiles/localframe"><img src="https://agentmods.dev/badge/skills/koolamusic/claudefiles/localframe/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/koolamusic/claudefiles/localframe"><img src="https://agentmods.dev/badge/skills/koolamusic/claudefiles/localframe.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00123 | $0.02018 |
| Opus 5.5 | $0.00049 | $0.00807 |
| Sonnet 5.5 | $0.00025 | $0.00404 |
| Haiku 4.5 | $0.00012 | $0.00202 |
Grade A, and why
localframe scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
3. Dry-run every command. For a terminal take, run `bash demo.sh` until every `curl | jq` line works against a real response. A jq filter written from memory is often wrong, and a mid-take failure costs a full re-shoot. How it starts
The opening of the file, as written. The whole thing — 130 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Localframe
Two kinds of take, one method:
- Browser: Playwright drives headless Chromium at 1440×900 through a paced walkthrough and records it.
- Terminal: a shell shot list types each command, runs it for real, and asciinema records the session.
The owner narrates over the result or uploads it as is. Takes are silent.
Templates and scripts live beside this file, deployed at ~/.claude/skills/localframe/. Call scripts by that full path and run them from the take directory:
| File | Use |
|---|---|
templates/walkthrough.mjs |
browser shot list with pause, shot, typeSlow, scrollSlow, waitFile |
templates/demo.sh |
terminal shot list with type_line, run, comment |
~/.claude/skills/localframe/scripts/doctor.sh [browser|terminal] |
pre-flight tool check; installs nothing |
~/.claude/skills/localframe/scripts/postflight.sh <take> ['<revoke check>'] |
the publish gate; exits 1 on any failure |
Copy a template into the take and edit it for the product. Don't build a shared library around it. Every product needs a different script.
Take layout
<take>/
walkthrough.mjs | demo.sh shot list, committed so the take can be reproduced
walkthrough.mp4 | demo.mp4 the video
raw/walkthrough.webm browser original
demo.cast, demo.gif terminal originals
01-sign-in.png 02-overview.png ... one numbered still per screen; also the shot list
listing.md one paragraph describing the recording
.take/ scratch, gitignored, mode 700: secrets.env, relay files
.take/secrets.env is the one place secrets live during a take. Set it to mode 0600. It holds BASE_URL, DEMO_KEY and anything else the shot list needs. The browser side writes a key it creates there. The terminal side sources it. Nothing secret goes in a command-line argument or in a committed script. Delete .take/ once postflight passes.
Pre-flight
- Run
~/.claude/skills/localframe/scripts/doctor.sh browserorterminalfrom the take directory. Install what it reports missing. Its hints are exact. Note that theaggcrate on crates.io is unrelated; installaggfrom asciinema's repo. - Create a demo key for this take. Never load a production key or a shared key into a recorded session. Give anything the walkthrough creates an obviously disposable name (
demo-take-2026-10-10). - Dry-run every command. For a terminal take, run
bash demo.shuntil everycurl | jqline works against a real response. A jq filter written from memory is often wrong, and a mid-take failure costs a full re-shoot. - Count the requests the shot list makes and keep them under the API's per-minute quota, so a retake doesn't hit a 429 mid-scene.
- Create the scratch directory with
mkdir -m 700 .take(the browser template also creates it), and add.take/to the repository's.gitignoreif the take lives in a repo.
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 130 lines · 123 tokens per session scan A e2b7d09fde5e
localframe is a skill published in the GitHub repository koolamusic/claudefiles (130 stars, last pushed 4d ago), licensed MIT. It adds 123 tokens to every session and 2,018 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-11.
Other skills, from other repositories
debug-optimize-lcp
Guides debugging and optimizing Largest Contentful Paint (LCP) using Chrome DevTools MCP tools. Use this skill whenever the user asks about LCP performance, slow page loads, Core Web Vitals optimization, or wants to understand why their page's main content takes too long to appear. Also use when the user mentions…
interactive-login
Log a CLI in with a browser/device-code flow (aws sso, gh, glab, gcloud, …) and save the result to the keychain so later commands can request it through execute.credentials.
pinchtab-mcp
Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.
azure-messaging-webpubsub-java
Build real-time web applications with Azure Web PubSub SDK for Java. Use when implementing WebSocket-based messaging, live updates, chat applications, or server-to-client push notifications.
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
open-design-browser
Inspect and exercise local HTML design artifacts with the existing agent-browser CLI, including screenshots and keyboard or interaction checks. Use when rendered evidence helps author or review a design; excludes browser installation, remote services, and unrelated browser automation.